A multi-depth camera capture system for point cloud library

The area of Human Computer Interaction (HCI) benefits largely from the release of low cost depth cameras such as Microsoft´s KINECT or ASUS´ XTION Pro. Depth, RGB or infrared image data are provided via a USB-2.0-connection. However, due to limited USB data rate, HCI applications usually cannot use more than two cameras per PC. To overcome this problem we propose a system which uses a client-server-based capture system. Servers send their data to a client, where it can be further processed. The proposed system based on the Point Cloud Library (PCL), where point clouds are computed and merged into a global registered point cloud. This application can be used in conjunction with an infrared tracking system to evaluate human pose or hand pose estimation.